1. Field
The present disclosure relates to computer systems and methods in which data resources are shared among data consumers while preserving data integrity and consistency relative to each consumer. More particularly, the disclosure concerns a mutual exclusion mechanism known as “read-copy update.”
2. Description of the Prior Art
By way of background, read-copy update (also known as “RCU”) is a mutual exclusion technique that permits shared data to be accessed for reading without the use of locks, writes to shared memory, memory barriers, atomic instructions, or other computationally expensive synchronization mechanisms, while still permitting the data to be updated (modify, delete, insert, etc.) concurrently. The technique is well suited to both uniprocessor and multiprocessor computing environments wherein the number of read operations (readers) accessing a shared data set is large in comparison to the number of update operations (updaters), and wherein the overhead cost of employing other mutual exclusion techniques (such as locks) for each read operation would be high. By way of example, a network routing table that is updated at most once every few minutes but searched many thousands of times per second is a case where read-side lock acquisition would be quite burdensome.
The read-copy update technique implements data updates in two phases. In the first (initial update) phase, the actual data update is carried out in a manner that temporarily preserves two views of the data being updated. One view is the old (pre-update) data state that is maintained for the benefit of read operations that may have been referencing the data concurrently with the update. The other view is the new (post-update) data state that is seen by operations that access the data following the update. In the second (deferred update) phase, the old data state is removed following a “grace period” that is long enough to ensure that the first group of read operations will no longer maintain references to the pre-update data. The second-phase update operation typically comprises freeing a stale data element to reclaim its memory. In certain RCU implementations, the second-phase update operation may comprise something else, such as changing an operational state according to the first-phase update.
FIGS. 1A-1D illustrate the use of read-copy update to modify a data element B in a group of data elements A, B and C. The data elements A, B, and C are arranged in a singly-linked list that is traversed in acyclic fashion, with each element containing a pointer to a next element in the list (or a NULL pointer for the last element) in addition to storing some item of data. A global pointer (not shown) is assumed to point to data element A, the first member of the list. Persons skilled in the art will appreciate that the data elements A, B and C can be implemented using any of a variety of conventional programming constructs, including but not limited to, data structures defined by C-language “struct” variables. Moreover, the list itself is a type of data structure.
It is assumed that the data element list of FIGS. 1A-1D is traversed (without locking) by multiple readers and occasionally updated by updaters that delete, insert or modify data elements in the list. In FIG. 1A, the data element B is being referenced by a reader r1, as shown by the vertical arrow below the data element. In FIG. 1B, an updater u1 wishes to update the linked list by modifying data element B. Instead of simply updating this data element without regard to the fact that r1 is referencing it (which might crash r1), u1 preserves B while generating an updated version thereof (shown in FIG. 1C as data element B′) and inserting it into the linked list. This is done by u1 acquiring an appropriate lock (to exclude other updaters), allocating new memory for B′, copying the contents of B to B′, modifying B′ as needed, updating the pointer from A to B so that it points to B′, and releasing the lock. In current versions of the Linux® kernel, pointer updates performed by updaters can be implemented using the rcu_assign_pointer ( ) primitive. As an alternative to locking during the update operation, other techniques such as non-blocking synchronization or a designated update thread could be used to serialize data updates. All subsequent (post update) readers that traverse the linked list, such as the reader r2, will see the effect of the update operation by encountering B′ as they dereference B's pointer. On the other hand, the old reader r1 will be unaffected because the original version of B and its pointer to C are retained. Although r1 will now be reading stale data, there are many cases where this can be tolerated, such as when data elements track the state of components external to the computer system (e.g., network connectivity) and must tolerate old data because of communication delays. In current versions of the Linux® kernel, pointer dereferences performed by readers can be implemented using the rcu_dereference ( ) primitive.
At some subsequent time following the update, r1 will have continued its traversal of the linked list and moved its reference off of B. In addition, there will be a time at which no other reader task is entitled to access B. It is at this point, representing an expiration of the grace period referred to above, that u1 can free B, as shown in FIG. 1D.
FIGS. 2A-2C illustrate the use of read-copy update to delete a data element B in a singly-linked list of data elements A, B and C. As shown in FIG. 2A, a reader r1 is assumed be currently referencing B and an updater u1 wishes to delete B. As shown in FIG. 2B, the updater u1 updates the pointer from A to B so that A now points to C. In this way, r1 is not disturbed but a subsequent reader r2 sees the effect of the deletion. As shown in FIG. 2C, r1 will subsequently move its reference off of B, allowing B to be freed following the expiration of a grace period.
In the context of the read-copy update mechanism, a grace period represents the point at which all running tasks (e.g., processes, threads or other work) having access to a data element guarded by read-copy update have passed through a “quiescent state” in which they can no longer maintain references to the data element, assert locks thereon, or make any assumptions about data element state. By convention, for operating system kernel code paths, a context switch, an idle loop, and user mode execution all represent quiescent states for any given CPU running non-preemptible code (as can other operations that will not be listed here). The reason for this is that a non-preemptible kernel will always complete a particular operation (e.g., servicing a system call while running in process context) prior to a context switch.
In FIG. 3, four tasks 0, 1, 2, and 3 running on four separate CPUs are shown to pass periodically through quiescent states (represented by the vertical bars). The grace period (shown by the dotted vertical lines) encompasses the time frame in which all four tasks that began before the start of the grace period have passed through one quiescent state. If the four tasks 0, 1, 2, and 3 were reader tasks traversing the linked lists of FIGS. 1A-1D or FIGS. 2A-2C, none of these tasks having reference to the old data element B prior to the grace period could maintain a reference thereto following the grace period. All post grace period searches conducted by these tasks would bypass B by following the updated pointers created by the updater.
Grace periods may synchronous or asynchronous. According to the synchronous technique, an updater performs the first phase update operation, invokes a function such as synchronize_rcu ( ) to await a grace period, then blocks (waits) until a grace period has completed, and then implements the second phase update operation, such as by removing stale data. According to the asynchronous technique, an updater performs the first phase update operation, specifies the second phase update operation as a callback, invokes a function such as call_rcu ( ) to await a grace period and invoke callback processing, then resumes with the knowledge that the callback will eventually be processed at the end of the grace period. Advantageously, callbacks requested by one or more updaters can be batched (e.g., on callback lists) and processed as a group at the end of an asynchronous grace period. This allows asynchronous grace period overhead to be amortized over plural deferred update operations.
Modern operating system kernels have been adapted to account for CPU low power states (such as, on Intel® processors, the C1E halt state, or the C2 or deeper halt states). Such operating system kernels can take advantage of low power state capabilities by using mechanisms that withhold regular timer interrupts from CPUs unless the CPUs need such interrupts to perform work. The dynamic tick framework (also called “dyntick” or “nohz”) available in current versions of the Linux® kernel is one such mechanism. Selecting the CONFIG_NO_HZ_FULL_IDLE compiler option provides nohz functionality for idle CPUs. In such Linux® kernel builds, the kernel's RCU grace period detection mechanism does not interfere with a CPU entering or remaining in a low power state, provided RCU does not require the CPU to acknowledge a quiescent state in order to end a grace period. The reason for this is because grace period detection processing is normally driven by the scheduling clock interrupt, and such processing cannot be performed on a CPU whose scheduling clock tick is suppressed.
Another Linux® kernel compiler option known as CONFIG_NO_HZ_FULL is used to withhold timer ticks from certain CPUs during times when they have only one runnable task. Such CPUs may be referred to as “adaptive-tick” or “tickless” CPUs. They may be designated at system boot time with the “nohz_full=” boot parameter. When one or more CPUs in a computer system are designated as a tickless CPU, the operating system kernel will avoid disturbing those CPUs whenever they have only a single runnable task. This option may be used to minimize OS jitter for HPC workloads or to improve latencies for real-time workloads. It is desirable that tickless CPUs executing a single runnable task in user mode should not be disturbed, either by scheduling-clock interrupts, IPIs, device interrupts, or kernel threads. All such potential sources of latency and jitter may be kept on some other CPU that has been designated for that purpose. For example, it is customary to exclude the boot CPU (CPU 0) from the nohz_full=set, so that this CPU always receives regular timer ticks and can be used to handle various timer-driven events.
More recently, a flavor of RCU named TASKS_RCU has been developed to allow freeing of “trampolines” (indirect jump vectors), which are used by some forms of tracing and also by some types of no-downtime live-kernel-patching mechanisms. TASKS_RCU is analogous to other flavors of RCU, but with a key difference being that grace period detection is handled by tracking when tasks (rather than CPUs) can no longer hold references to shared data. Other differences include the fact that readers do not invoke any RCU read-side primitives, such as the rcu_read_lock ( ) or rcu_read_unlock ( ) functions used in other flavors of RCU. For updaters, a call_rcu ( ) variant named call_tasks_rcu ( ) is provided for posting callbacks to be processed at the end of a TASKS_RCU grace period.
TASKS_RCU uses voluntary context switch (not preemption) and user mode execution as its sole quiescent states (possibly with addition of the idle loop). At the beginning of a TASKS_RCU grace period, the TASKS_RCU grace period detection mechanism creates a separate list of task structures corresponding to every runnable process in the system. Tasks that are not runnable are deemed unable to hold a reference to data structures protected by TASKS_RCU, and are not therefore included. Each task on the runnable holdout list is deemed a “holdout task,” and a special “rcu_tasks_holdout” flag is set in its task structure at the beginning of the TASKS_RCU grace period. This flag is cleared by the operating system scheduler whenever the task voluntarily relinquishes the CPU or returns to user mode. A TASKS_RCU kernel thread loops through the list of holdout tasks on the runnable task list, and removes any that have had their rcu_tasks_holdout flag reset. Once the runnable list is empty, callbacks posted prior to the start of the current grace period, e.g., using the call_rcu_tasks ( ) function, may be invoked.
Unfortunately, the above-described TASKS_RCU grace period detection mechanism will ignore a tickless CPU running a single task in tickless user mode at the time the task is added to the runnable task list. While the task is running, there would be neither a context switch nor a scheduling-clock interrupt to tell TASKS_RCU that the task in question has passed through a quiescent state. Instead, TASKS_RCU would have to wait for the next voluntary context switch, which could extend the grace period indefinitely.
It would therefore be desirable for TASKS_RCU to be able to detect which tasks are running in tickless user mode on tickless CPUs, without disturbing these CPUs.